A Comprehensive Look at Mainstream Chip Solutions for Smart Glasses

AI glasses and AR glasses are evolving from products that “can listen, take photos and answer questions” toward devices that “can perceive scenes, display content and understand the environment”. In this evolution, chips are no longer simple Bluetooth audio main controllers. Instead, they directly determine the shooting image quality, AI response speed, battery life, heat dissipation, weight, display capability and overall cost of the glasses.

Photo taken by Aibang at the Yijing Virtual booth during the Shenzhen Smart Glasses Exhibition on June 16.

Unlike smartphones and tablets, all core hardware of smart glasses is packed into the temple and frame with extremely limited space. While retaining the wearing form of ordinary glasses, they need to accommodate cameras, microphones, speakers, batteries, antennas, sensors, display modules, and even optical waveguides, Micro LED optical engines and other components.

This makes chip selection extremely critical:

• Low power consumption Smart glasses are equipped with small-capacity batteries. Continuous high-load operation of chips will easily lead to short battery life and hot temples.
• Powerful ISP performance For AI glasses with cameras, photography, video recording, image stabilization, night shooting, HDR and distortion correction all rely on ISP image processing capabilities.
• Sufficient AI capability Voice wake-up, noise reduction, translation, object recognition, text summarization, visual question answering and other functions require collaboration between on-device AI and cloud large models.
• Stable connectivity Glasses need to link with mobile phones, cloud servers, earphones, watches and other devices. Wi-Fi, Bluetooth and low-latency transmission are of great importance.
• Additional computing power for AR glasses with displays If the glasses feature a display, they also need to support graphics rendering, spatial positioning, display driving, 3DoF/6DoF, SLAM and other capabilities.

Therefore, the chip roadmap for AI/AR glasses is essentially a continuous trade-off among performance, power consumption, size and cost.

Most chips used for smart glasses today are not custom-designed for glasses. They are migrated from smartphones, smartwatches and other fields. Take Qualcomm AR1 as an example. It is one of the well-rounded chip solutions for AI glasses currently available. The AR1 boasts complete image, AI and connectivity capabilities. However, not all its functions can be fully utilized in AI glasses, which may result in redundant features, higher costs and greater pressure on power consumption.

There is no single chip solution that can dominate all smart glasses products. Manufacturers select different chip combinations according to product positioning, battery life requirements, cost budget and mass production difficulty.

Below we will introduce common solutions, representative chips, suppliers and corresponding smart glasses products in the market.

Three Mainstream Chip Solutions for Smart Glasses

1. System-on-Chip (SoC) Solution

This is the most typical solution for AI glasses and some AR glasses. It integrates CPU, GPU, NPU/DSP, ISP, Wi-Fi, Bluetooth and other functions into a single main SoC.

Disassembly of Oakley x Meta HSTN smart glasses equipped with Qualcomm AR1. Source: IFIXIT, Author: Elizabeth Chamberlain.

Representative Chips: Qualcomm AR1, Unisoc W517, etc. Advantages: High integration, strong performance and relatively complete development ecosystem, suitable for photography, AI recognition, voice interaction and lightweight AR display. Disadvantages: Higher cost, obvious pressure on power consumption and heat dissipation. Representative Glasses: Oakley x Meta HSTN; INMO Air2.

From the market perspective, Qualcomm AR1 is one of the most iconic chips for AI glasses.

2. Dual-chip Solution: SoC + Low-Power Co-processor

As AI glasses require long standby time, voice wake-up, music playback, calls and translation, continuous operation of a single high-performance SoC will bring obvious power consumption pressure. Therefore, many brands adopt the “dual-chip architecture”.

The working principle is straightforward: the high-performance SoC handles shooting, visual AI and complex computing; the low-power co-processor takes charge of voice wake-up, Bluetooth audio, sensor monitoring and lightweight tasks. When users are not taking photos or interacting with large models, the main chip can enter sleep mode, while the low-power chip maintains basic functions.

Representative Glasses: Xiaomi AI Glasses (Qualcomm AR1 + BES2700H), Thunderbird V4 (Qualcomm AR1 + BES2800), Rokid Glasses (Qualcomm AR1 + NXP RT600).

3. MCU-grade SoC + ISP Solution

The third solution focuses more on cost and power consumption control. It usually adopts a low-power MCU or wearable SoC as the main controller, with an external independent ISP chip to process camera images.

Infinix AI Glasses Pro. Source: Infinix. Li Auto Livis AI Glasses. Source: Li Auto.

This solution is suitable for AI glasses mainly featuring photo & video recording, voice assistant, translation and meeting recording. Advantages: Low cost and longer battery life. Disadvantages: Inferior overall performance, platform ecosystem and complex AI capability compared with complete SoC solutions such as Qualcomm AR1.

Representative Chips & Solutions: BES2800 + external ISP, ISP solutions from StarSoc, EZMicro, Fullhan Micro, Allwinner V821, Actions ATS308X series, etc. Representative Products: Infinix AI Glasses Pro, Li Auto Livis AI Glasses, Qingcheng Wireless AI Glasses, MLIB Magic Treasure AI Glasses.

This type of solution carries great market significance. If AI glasses aim for mass adoption, they cannot remain in the price range of more than 2,000 to 3,000 RMB. Combinations of domestic low-power SoC, Bluetooth audio chips, ISP chips and AIoT chips have the potential to push AI glasses to lower price brackets.

SoC Suppliers

1. Qualcomm

Qualcomm is one of the most representative chip suppliers for AI and AR glasses. It excels in image processing, AI, wireless connectivity and mature XR ecosystem, targeting mid-to-high-end AI glasses, lightweight AR glasses and camera-enabled smart glasses.

2. BES Technology

BES Technology is a key domestic supplier of low-power co-processing solutions for AI glasses. Its chips are often paired with Qualcomm AR1 to form a dual-chip architecture. Qualcomm handles imaging, AI and complex computing, while BES takes charge of Bluetooth audio, voice wake-up, low-power connectivity and lightweight tasks. Xiaomi AI Glasses uses Qualcomm AR1 together with BES2700H Bluetooth audio processor. Thunderbird V4 also adopts Qualcomm AR1 + BES2800 dual-chip architecture.

The BES2800 series is not merely a co-processor. It itself is a low-power wearable SoC for AI glasses, TWS earbuds and smartwatches, equipped with audio processing, wireless connectivity, on-device AI and system control capabilities. For some mid & low-power AI glasses solutions, BES2800 can also act as the main SoC, paired with an external ISP or visual processing chip to complete photography, video recording and image optimization.

3. Unisoc

Unisoc mainly provides domestic wearable SoC solutions, featuring localization, low power consumption, cost control and good adaptability for wearable platforms. Compared with Qualcomm AR1, Unisoc solutions are more suitable for translation glasses, lightweight display glasses and domestic AI wearable platforms.

Representative chips include W517 and W337. W517 is positioned as a 4G AI wearable platform with high performance, low power consumption and high integration. INMO Air2 and INMO GO2 once adopted the W517 solution; INMO GO3 uses W337 paired with WQ7036AX from WiMi.

4. Rockchip

Rockchip has two product lines for AI/AR glasses: RK series general SoCs, mostly used for AR glasses boxes, external computing boxes, video glasses hosts and edge AI computing platforms; RV series visual SoC/ISP chips, which can be embedded inside AI camera glasses for camera image processing, photo & video recording, HDR, image stabilization, low-light enhancement and low-power visual processing.

Representative chips: RK3566, RK3588, RV1106B, RV1103B, RV1126B, etc.

5. Allwinner Technology

Allwinner is a noteworthy domestic supplier of low-cost AI glasses solutions. Its focus is not high-end AR computing, but low-power, low-cost and quickly mass-producible AI camera glasses.

Allwinner V821. Source: Allwinner Technology.

The representative chip V821 is a dual RISC-V architecture video SoC integrating high-performance ISP and hardware encoding unit, supporting 4MP camera input and H.264 encoding, with built-in Wi-Fi and Audio Codec. Products adopting V821 include Qingcheng Wireless AI Glasses, MLIB Magic Treasure AI Glasses and some Huaqiangbei AI camera glasses.

6. Ingenic

Ingenic mainly develops low-power video SoC and ISP visual processing chips. It is not a full-featured AI glasses SoC like Qualcomm AR1, but ideal for camera-focused AI glasses, focusing on photography, video recording, fast startup, low power consumption and image processing.

For example, KANAAN K1 AI Glasses uses Ingenic T31ZX/C100 related solution to realize quick snapshot and video recording.

7. MediaTek

MediaTek provides main SoC for AI glasses, responsible for relatively complex processing tasks in the whole device, including camera image capture, video processing, system operation, AI algorithm invocation and multimedia support. Take Shenzhen Xingyi Intelligent K900 series as an example, it adopts multi-chip collaborative architecture of MTK + BES2700. MTK handles video and image processing, BES2700 takes care of Bluetooth audio and voice interaction, and an extra low-power voice wake-up chip is deployed for standby wake-up and voice recognition.

8. Hexagon Semiconductor

Tianxiang HX77 series is a dedicated low-power image/display processing SoC for AI+AR glasses built on RISC-V architecture, integrating CPU, GPU, DPU, PMU and multiple display interfaces. It mainly undertakes image display processing, low-latency rendering, 3DoF spatial hovering and dual-screen display in AR glasses. It features low power consumption and high integration with outstanding display processing capability, yet it does not focus on heavy on-device AI inference; complex AI tasks mostly rely on mobile phones or cloud collaboration.

Photo taken by Aibang at Hexagon Semiconductor booth during the Shenzhen Smart Glasses Exhibition on June 16.

MCU / Low-Power Co-processing Chips

1. WiMi Microelectronics

WiMi Microelectronics focuses on audio, connectivity and low-power voice interaction for AI glasses. Representative chip: WQ7036 series.

NIMO Smart Glasses. Source: WiMi Microelectronics.

The WQ7036 series has been applied in Looktech AI Glasses, KANNAN-K2, NIMO AI Glasses, Xgimi Memo and INMO GO3. It handles voice processing, Bluetooth connection, audio transmission, multi-microphone noise reduction and AI wake-up.

2. NXP

NXP acts as a low-power co-processor for AI glasses, represented by RT600 series. These chips are suitable for always-on tasks such as voice wake-up, sensor monitoring and low-power task scheduling, so the high-performance main SoC does not need to run continuously to reduce overall power consumptio

Rokid Glasses is a typical case adopting Qualcomm AR1 + NXP RT600 dual-chip solution: AR1 is responsible for AI and image processing, while RT600 handles low-power voice wake-up and task management.

3. Actions Technology

Actions Technology’s strengths lie in low-power wireless audio, Bluetooth connectivity, AI noise reduction and wearable SoC. Its chips can serve as the main controller for lightweight AI glasses or audio/connectivity co-processor. Representative chips: ATS308X series, ATW609X series. Multiple companies including INMO, Halliday and Xingyi Intelligent have released AI glasses solutions based on Actions ATS308X.

4. StarSoc Technology

StarSoc Technology is a key supplier of ISP/visual processing chips for AI glasses. Its representative chip SSC309QL targets small wearable devices like smart glasses, highlighting low power consumption, small size, ISP image processing and AI image enhancement. Take KANNAN K2 as an example, it uses the combination of StarSoc SSC309QL + WiMi WQ7036. SSC309QL takes charge of camera image processing, photo & video recording and AI image enhancement.

5. EZMicro

EZMicro provides independent ISP capability for AI glasses. Paired with low-power main controllers such as BES2800, it realizes photography, video recording, noise reduction, HDR, white balance, sharpening and other image processing functions. A typical solution is BES2800 + EZMicro SA62105X. BES2800 handles system control, audio, Bluetooth connection, voice wake-up and lightweight AI tasks; SA62105X takes charge of camera ISP, photo & video recording and image processing. Li Auto Livis AI Glasses and L’Atitude 52°N adopt this solution.

6. Fullhan Microelectronics

Fullhan Microelectronics’ representative chip MC6350 is a 12MP ISP chip for low-power, miniaturized devices such as smart glasses. Adopting 12nm low-power process with a package size of approximately 8×8mm and integrated DDR, it helps reduce BOM cost and PCB footprint. For AI glasses, the core value of MC6350 is to improve shooting quality, especially imaging in low-light environments.

For instance, Kuangming Intelligent launched a dual-chip AI smart glasses solution of BES2800 + Fullhan MC6350.

7. CoreChips Microelectronics

CoreChips Microelectronics is a domestic SoC supplier newly entering the AI glasses chip track, targeting smart glasses with cameras, displays and on-device AI capabilities. Its representative chip ARS45 is an exclusive SoC for AI glasses, emphasizing low power consumption, small package, on-device AI, ISP image processing and display capability. iFLYTEK AI Translation Glasses uses BES Technology BES2800 + CoreChips ARS45 chip solution.

8. Jieli Technology

Jieli Technology specializes in Bluetooth audio SoC, wireless connectivity and low-cost consumer-grade chips. In AI glasses, it mostly works as an audio connectivity chip or low-power co-processor for open audio, Bluetooth connection, voice interaction and call noise reduction. Jieli’s JL7018F6, AC7018 and other Bluetooth audio SoCs can form a “main controller + audio” dual-chip solution together with visual main controllers such as Allwinner V821. For example, ANTAVIC CUKE M02 adopts “JL7018F + V821” dual-chip architecture.

In addition to the above enterprises, Bluetrum, Telink, Espressif, Realtek and Airoha also make layout in the field of low-power chips for AI glasses.

Judging from current products, AI glasses and AR glasses will not be dominated by a single chip route. Multiple solutions will coexist according to product positioning. However, no matter which solution is adopted, the core challenge remains unchanged: how to achieve stronger computing power, lower power consumption, more stable connectivity and a more natural interactive experience within extremely limited space.